The intersection of artificial intelligence (AI) and various industries is transforming the way we live and work. This week, we saw significant developments in AI, from licensing and training to security and innovation. Here's a summary of the key news and trends:
Technical Needs to Establish Licensing for AI Training
Despite the still unsettled legal environment, 2024 saw the first stirrings of a between AI companies and rights owners for the use of copyrighted works to train generative AI models. Nearly three dozen content licensing deals were announced or reported by the end of 2024 involving rights owners in the U.S., Canada, and Europe. Most of the content licensing deals announced so far have had a in common: direct, one-to-one deals, limited windows, and unlimited use. For a more robust, sustainable, and to evolve, at least three key technical capabilities will need to be in place: a machine-readable means to unambiguously identify a piece of content, tools that effectively control access to and collection of copyrighted content at the asset level, and the means to track, measure, and properly attribute the use of a piece of content by and within an AI system.
Thousands of Live Credentials Found in AI Training Data
A popular open dataset used to train Large Language Models (LLMs) is riddled with active credentials, including live API keys and passwords. Truffle Security found 11,908 hardcoded API keys and passwords in the dataset. This practice can lead to the deterioration of guardrails, as models trained on exposed APIs start to assume that this is a common practice and start to repeat the activity.
DeepSeek Open-Sources AI Model Training Details
Chinese AI start-up DeepSeek has released a series of open-source projects on GitHub, revealing details about how it trained its low-cost, high-performance models that shocked international markets earlier this year. The company released two models in January, the V3 large language model and the R1 reasoning model, which it said achieved performance on par with major Western competitors but were trained for a fraction of the cost.
How AI and Security Are Reshaping DevSecOps and DevOps Tooling
The convergence of application development, artificial intelligence, and security is transforming how organizations build and maintain software, driving the rise of DevSecOps. AI-driven automation is reshaping software development, but it comes with risks. While organizations are eager to use AI for efficiency, security remains a key concern.
Google's Gemini Now Lets You Ask Questions Using Videos or What's on Your Screen
Google is adding new features to its AI assistant, Gemini, that let users ask it questions using video or content on the screen in real time. The company showed off a new “Screenshare” feature, which lets users share what’s on their phone’s screen with Gemini and ask questions about the company.
Guggenheim and Legendary Billionaires Create $40bn Pot to Make AI Bets
Mark Walter and Thomas Tull want to use artificial intelligence to spur innovation in their finance, sport, and media holdings. They have created a $40 billion pot to make AI bets.
US Suspicion May Be Right, Singapore May Have an 'Nvidia-DeepSeek Problem'
Singapore Police have arrested three men for fraud linked to illegal re-export of Nvidia GPUs to Chinese AI company DeepSeek, bypassing US trade sanctions.
Can AI Outperform Doctors in Diagnosing Infectious Diseases?
Artificial Intelligence (AI) is revolutionizing diagnostics, improving disease detection and treatment planning. However, navigating ethical and data integration challenges is crucial for its successful implementation. Recent studies have shown that AI algorithms can outperform doctors in diagnosing certain infectious diseases, such as pneumonia and tuberculosis.
Varonis at the 2025 Gartner Security & Risk Management Summit: AI Innovation and Data Security
Varonis will be attending the Gartner Security & Risk Management Summit, March 10-11 in Mumbai, India. The company will be showcasing its latest cloud security offerings and exploring the importance of implementing proactive data security for AI.
Key Takeaways
- AI licensing and training are becoming increasingly important, but technical capabilities are still lacking.
- AI training data is often riddled with active credentials, which can lead to security risks.
- Open-sourcing AI model training details can help improve transparency and efficiency.
- AI-driven automation is reshaping software development, but security remains a key concern.
- Google's Gemini is adding new features to let users ask questions using video or content on the screen in real time.
- AI is being used to spur innovation in various industries, including finance, sport, and media.
- AI can outperform doctors in diagnosing certain infectious diseases, but ethical and data integration challenges must be addressed.
- Varonis is showcasing its latest cloud security offerings and exploring the importance of implementing proactive data security for AI.
Sources
- Technical Needs to Establish Licensing for AI Training
- Thousands of live credentials found in AI training data
- DeepSeek Open-Sources AI Model Training Details
- How AI and security are reshaping DevSecOps and DevOps tooling
- MWC 2025 Live Blog: Phones, Wearables, Robots and AI Live From Barcelona
- Google's Gemini now lets you ask questions using videos or what's on your screen
- Guggenheim and Legendary billionaires create $40bn pot to make AI bets
- US suspicion may be right, Singapore may have an 'Nvidia-DeepSeek problem' - The Times of India
- Can AI Outperform Doctors in Diagnosing Infectious Diseases?
- Varonis at the 2025 Gartner® Security & Risk Management Summit: AI Innovation and Data Security